DOI Seite / Zitierlink:
https://doi.org/10.11588/diglit.45927#0057
3.6 The Roofing Problem
55
ever constructed under a single roof. Severely damaged by fire in A.D. 80,120 it remained “open to the sky” in
Dio’s time “since it [the roof] could not be put together again,”121 so we must not imagine that such technical
feats were a commonplace during the Empire, even in Rome itself. But enormous timbers of larch, fir and pine
did exist which clearly inspired awe and might well have been procured for an important civic building in this
wealthy capital of Asia by a patron eager to demonstrate his generosity towards his city. Even if such long
beams were not available, shorter lengths could be joined together using carpentry techniques well known to
ancient builders.
The triangulated tie-beam trusses used by Roman builders were placed at intervals determined by specific
support conditions such as the location of columns, piers or buttressing. As each open frame had to support
tiles or other cladding in addition to its own weight, it depended on internal bracing consisting of smaller
timbers joined, strapped or doweled together to neutralize the various forces inherent in the system such as
compression, flexion (bending) and tension (pulling). In a typical truss the tie-beams, each made either from a
single timber or from two timbers joined and strapped together, were secured to profiled corbels set in the tops
of opposite walls. The feet of the sloping principal rafters were rabbeted into the ends of this beam and tied
together at the top. A vertical “kingpost” hung from the apex, was doweled to the tie-beam preventing it from
sagging. Drawings made of the great late-antique trusses of S. Paolo fuori le mura in Rome before they burned
in 1823 show the king-post clasped between doubled tie-beams,122 a system which Vitruvius describes for the
roof covering his own basilica at Fanum.123 Additional internal bracing timbers including horizontal, vertical
and diagonal beams were used in various combinations. A very old roof in the monastery of St. Catherine in
the Sinai,124 dating from the 6th century, uses “scissor braces,” in which diagonal struts connected at the bottom
to the tie-beam and king-post join the principal rafters at right angles near the top, a technique most suitable
for roofs of steep pitch. In the shallower roof hypothetically restored for the Bouleuterion at Aphrodisias, the
maximum span of 27 m is divided into four equal parts by a king-post and two queen-posts.125 The truss thus
formed is made more rigid by joining the tops of the queen-posts with a collar beam.
Trussed roofs utilized a system of secondary timbers which permitted the actual covering of the area over
which they were erected. A series of purlins, equally spaced, spanned the distance between the primary rafters,
supporting in turn common rafters of smaller cross-section set at right angles to them. The cladding rested
either on thin batons or on a “plating” made of boards. The type of cladding used played a part in determining
a roof’s design. Steep roofs, capable of bearing heavier weights, could be covered with metal sheeting which
was fixed to the timbers. Roof tiles depended on gravity and friction and could be used only on relatively shal-
low roofs with a pitch of less than about 20%. The great quantities of roof tiles found during excavation of the
Bouleuterion at Ephesos suggest that its roof belonged to the latter category.
The roofs of Basilical halls, the only type of large ancient roof we have real information about, consisted
of identical units which were multiplied along an axis as required. Trusses for such buildings were isosceles
triangles of uniform size composed of bracing members of standard arrangement and scantling. The class of
buildings to which our Bouleuterion belongs added a further design factor in that it required trusses of differ-
ent lengths. Furthermore, if all the trusses in a roof took the form of isosceles triangles, the ridge would have
curved both in plan and elevation, presenting a decidedly awkward appearance. R. Meinel has assumed that
only the longest central trusses had sides of equal length and that their apex determined the position of the
ridge which bisected the plan parallel to the building’s facade.126 This is not the only possibility, however, and
at Ephesos there is some evidence that another solution was employed.
A series of trusses would have had to be tied together at the top with short beams spanning the distances
between them to provide lateral stability, and then anchored in the walls at both sides by diagonal struts in the
same vertical plane. Assuming central trusses of isosceles pattern, such a roof ridge at Ephesos would have
been aligned not with buttresses but with sections of screen wall only 0.75 m thick, which would appear too
120 Dio 66, 24, 2.
121 Dio 55, 8, 4.
122 Meinel 1980, 343 fig. 144,4.
123 Vitr. 5, 1, 6-10.
124 Galey 2003, 44-47. 94-95.
125 Cf. the design of the Aphrodisias roof, in: Meinel 1980, 323-326 fig. 137.
126 Meinel 1980, 349.
55
ever constructed under a single roof. Severely damaged by fire in A.D. 80,120 it remained “open to the sky” in
Dio’s time “since it [the roof] could not be put together again,”121 so we must not imagine that such technical
feats were a commonplace during the Empire, even in Rome itself. But enormous timbers of larch, fir and pine
did exist which clearly inspired awe and might well have been procured for an important civic building in this
wealthy capital of Asia by a patron eager to demonstrate his generosity towards his city. Even if such long
beams were not available, shorter lengths could be joined together using carpentry techniques well known to
ancient builders.
The triangulated tie-beam trusses used by Roman builders were placed at intervals determined by specific
support conditions such as the location of columns, piers or buttressing. As each open frame had to support
tiles or other cladding in addition to its own weight, it depended on internal bracing consisting of smaller
timbers joined, strapped or doweled together to neutralize the various forces inherent in the system such as
compression, flexion (bending) and tension (pulling). In a typical truss the tie-beams, each made either from a
single timber or from two timbers joined and strapped together, were secured to profiled corbels set in the tops
of opposite walls. The feet of the sloping principal rafters were rabbeted into the ends of this beam and tied
together at the top. A vertical “kingpost” hung from the apex, was doweled to the tie-beam preventing it from
sagging. Drawings made of the great late-antique trusses of S. Paolo fuori le mura in Rome before they burned
in 1823 show the king-post clasped between doubled tie-beams,122 a system which Vitruvius describes for the
roof covering his own basilica at Fanum.123 Additional internal bracing timbers including horizontal, vertical
and diagonal beams were used in various combinations. A very old roof in the monastery of St. Catherine in
the Sinai,124 dating from the 6th century, uses “scissor braces,” in which diagonal struts connected at the bottom
to the tie-beam and king-post join the principal rafters at right angles near the top, a technique most suitable
for roofs of steep pitch. In the shallower roof hypothetically restored for the Bouleuterion at Aphrodisias, the
maximum span of 27 m is divided into four equal parts by a king-post and two queen-posts.125 The truss thus
formed is made more rigid by joining the tops of the queen-posts with a collar beam.
Trussed roofs utilized a system of secondary timbers which permitted the actual covering of the area over
which they were erected. A series of purlins, equally spaced, spanned the distance between the primary rafters,
supporting in turn common rafters of smaller cross-section set at right angles to them. The cladding rested
either on thin batons or on a “plating” made of boards. The type of cladding used played a part in determining
a roof’s design. Steep roofs, capable of bearing heavier weights, could be covered with metal sheeting which
was fixed to the timbers. Roof tiles depended on gravity and friction and could be used only on relatively shal-
low roofs with a pitch of less than about 20%. The great quantities of roof tiles found during excavation of the
Bouleuterion at Ephesos suggest that its roof belonged to the latter category.
The roofs of Basilical halls, the only type of large ancient roof we have real information about, consisted
of identical units which were multiplied along an axis as required. Trusses for such buildings were isosceles
triangles of uniform size composed of bracing members of standard arrangement and scantling. The class of
buildings to which our Bouleuterion belongs added a further design factor in that it required trusses of differ-
ent lengths. Furthermore, if all the trusses in a roof took the form of isosceles triangles, the ridge would have
curved both in plan and elevation, presenting a decidedly awkward appearance. R. Meinel has assumed that
only the longest central trusses had sides of equal length and that their apex determined the position of the
ridge which bisected the plan parallel to the building’s facade.126 This is not the only possibility, however, and
at Ephesos there is some evidence that another solution was employed.
A series of trusses would have had to be tied together at the top with short beams spanning the distances
between them to provide lateral stability, and then anchored in the walls at both sides by diagonal struts in the
same vertical plane. Assuming central trusses of isosceles pattern, such a roof ridge at Ephesos would have
been aligned not with buttresses but with sections of screen wall only 0.75 m thick, which would appear too
120 Dio 66, 24, 2.
121 Dio 55, 8, 4.
122 Meinel 1980, 343 fig. 144,4.
123 Vitr. 5, 1, 6-10.
124 Galey 2003, 44-47. 94-95.
125 Cf. the design of the Aphrodisias roof, in: Meinel 1980, 323-326 fig. 137.
126 Meinel 1980, 349.
